I want to evaluate a graph kernel designed for large graphs (> 10^6 nodes). Hence, I'm looking for suitable graph data sets, i.e., a set of (huge) graphs and corresponding classes.

Any ideas?

  • What type of graphs are you looking for? Would twitter data help?
    – sheß
    Sep 17 '15 at 12:38
  • @sheß: Both directed and undirected graphs are fine. Each graph should belong to a class so I can perform classification. Could you please elaborate on the twitter data. Sep 17 '15 at 13:06

Try Stanford Large Network Dataset Collection

https://snap.stanford.edu/data/egonets-Facebook.html[is facebook and has undirected graphs

Directed graphs twitter is listed undirected and directed graphs.

Hope it answer your question.
  • Thanks, I'm aware of SNAP. Unfortunately, most of the datasets consist of a single large graph. Sep 17 '15 at 16:16
  • Than will be need to build on your own. Check databricks graphX databricks-training.s3.amazonaws.com/…
    – n1tk
    Sep 17 '15 at 16:25
  • @Christopher. how large do you need?
    – n1tk
    Sep 17 '15 at 22:47
  • @sb709: Let's say 100 graphs for each class (two classes are sufficient), each graph should have at least 10^6 nodes. Sep 18 '15 at 13:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.